Determination of image resolution from SAS image statistics

The quality of a SAS image is affected by the coherence across the synthetic aperture which is directly related to errors in motion estimation, compensation, or physical phenomena (e.g. medium instability or acoustic multipath). Operational use of a SAS system can benefit from metrics that estimate image quality and identify areas where sonar imagery is suboptimal. Furthermore, an estimate of the coherence across the synthetic aperture would be necessary to maintain a system calibrated response and use acoustic measurements to determine target or seafloor physical properties. This paper proposes techniques to determine the mean point object response and relates this parameter to image resolution. Unlike previous attempts to determine image resolution that relied on the use of specifically designed targets intended to provide an ideal point response, this technique relies on a statistical approach where the image is assumed to contain many point-like objects. The point objects are identified by comparison with seafloor statistics, and the 3dB peak width is measured for all identified point objects in the image. Analysis of the point object results provides the mean image resolution in two dimensions as well as local variability of resolution across the image. The resolution response provided by this technique was verified by reprocessing samples of SAS imagery while varying the aperture length and bandwidth and determining the resolution of the sample images. Comparison of the predicted and measured resolution response is shown.